Skip to content
SatarovMOCR

Decorative 3D background: a mission control room. Tiered rows of consoles face a front wall carrying three projection boards — a ground track marking the cities where he has studied and worked, a strip chart of repositories created per year by domain, and a GO / STAND BY status board. Scrolling moves the viewpoint down the aisle toward the wall. Every value shown on the boards is also written out in the page text below.

Mansurbek Satarov

Trustworthy AI ResearcherML Systems EngineerFull-Stack Developer

Ph.D. researcher building trustworthy AI systems, studying how model behavior emerges during training, and engineering production software across machine learning, computer vision, agentic AI, cloud infrastructure, and interactive web applications.

Open to research collaborations and internships

The poll

Before a burn, a flight director calls every station and waits for an answer. This site answers the same way about its own evidence: a station reports GO only where a public, machine-checkable artifact backs the claim.

Flight: all stations, go / no-go for evaluation

Polling…

  1. Systems———20 of 22 case studies resolve to a public repository
  2. Network———45 repositories classified as own work from commit data
  3. Sim Bay———Five simulations, each with a pure core under unit test
  4. Science———Research direction stated; individual work not yet citable
  5. Declared———2 projects are résumé-declared with no public repository

Stand by is not a failure state — it marks work declared on a résumé that no public repository can confirm. It is on this board so you know exactly which claims you can check yourself.

Research systems, agentic pipelines and production ML, each with the architecture that made it work and the results it actually produced.

Flight data file — selected systems

  1. 001

    Multi-Agent Deep Research

    A LangGraph supervisor/worker research workflow that plans, gathers evidence, extracts structured findings, and writes sourced reports.

    Python · LangGraph · OpenAI API

    Complete
  2. 002

    Retail Intelligence Platform

    End-to-end Azure retail analytics: customer lifetime value, churn risk, basket associations, household search, and cached dashboards.

    FastAPI · React · Vite

    Complete
  3. 003

    RAG Assistant

    A modular retrieval-augmented generation pipeline: ingestion, chunking, embeddings, ChromaDB vector search, and a conversational interface.

    Python · Flask · ChromaDB

    Complete
  4. 004

    DinoMind Evolution

    NEAT neuroevolution learns to play a dino runner — neural networks evolve generation by generation to dodge obstacles.

    Python · Pygame · NEAT-Python

    Complete
  5. 005

    DocuParse

    High-performance PDF/EPUB/MOBI → structured Markdown conversion with AI layout detection, LaTeX equation handling, and multilingual support.

    Python · OCR · Layout-detection models

    Complete
  6. 006

    DigitRecognizer

    A neural network handwritten from scratch in NumPy — forward propagation, backprop, and gradient descent on MNIST, no frameworks.

    Python · NumPy · Matplotlib

    Complete

Ph.D. research on how model behavior forms during training, and how to keep it predictable once it has.

Channel select

Channel readout

Trustworthy machine learning

Making model behavior predictable, auditable, and safe to depend on.

1 / 10 channels

Every public repository, labeled by a relationship computed from commit data. Forks and stars are listed as exploration and reading — they are never counted as contribution.

Own work
45
Forks kept for study
30
Saved references
68
Technologies used
59

Own work by domain

  • Web20
  • Data science7
  • Computer vision6
  • LLM & agents4
  • Systems2
  • Cloud2
  • Games2
  • Other1
  • Machine learning1

Bars count repositories the classifier resolved to his own commit history. Forks and stars are excluded from every bar on this panel.

Label definitions

Built by me
Owned, not a fork, with a substantial own-commit history.
Co-built
Owned, with two or more active co-contributors.
Contributed
Someone else's repository, with merged pull requests.
Forked & explored
A fork kept for study. Counts as reading, not work.
Studied / used
Depended on or read closely, without commits.
Saved resource
Starred for reference. Zero contribution claimed.

Five simulations that run in the browser. Each has a pure, deterministic core under unit test — the same code the tests replay is the code you play.

Capcom

The only station that talks back. Open to research collaboration, Ph.D.-adjacent work, and engineering that has to hold up.

Open a channelsataromk@mail.uc.edu

Cincinnati, OH
Replies usually within two days.